I think by " higher-resolution space" you maybe mean transformed into native
space. I guess you are concerned because we often down sample to 2x2x2
voxels, meaning there are more voxels in the analysis.
However, if there was any real advantage to stating in lower-res voxel
space, we would all be doing that. Think of it this way: its not like the
same number of voxels will be significant at 2x2x2 as at 4x4x4. If we have
200 mm2 of brain activity, we will tend to see that as 3 voxels at 4x4x4 and
25 voxels at 2x2x2. So while the voxel sizes are smaller, and the corrected
extend threshold becomes larger, the clusters should also increase in size.
You can do as you propose, but I do not believe it will convey any advantage
to you, statistically. However, you are welcome to try it and see what
happens.
best of luck,
Colin Hawco, PhD
Neuranalysis Consulting
Neuroimaging analysis and consultation
www.neuranalysis.com
[log in to unmask]
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
Behalf Of Karolis, Slava
Sent: October-03-17 9:15 AM
To: [log in to unmask]
Subject: [SPM] p-values in higher-resolution space
Dear all,
Will an SPM analysis in a higher-resulution space be less likely to produce
significant results due to a greater number voxels (=number of multiple
comparisons)?
My situation is as follows: I am doing 1st level in a native fMRI space and
then co-register beta maps to a higher resolution study-specific template
for a group analysis.
Best Wishes,
Slava
|